pushdown?

[tooptoop4]

if i have a large data source like redshift (or any data source) that i want to profile/run expectations against, is the behavior like method 1 or method 2?

method 1:
great_expectations generates DISTINCT/MAX/MIN..etc queries and runs against redshift, so all the compute is pushed down to redshift and only the results are returned to python

method 2:
great_expectations runs select *, returns all data (+millions of rows) into a python object, then runs DISTINCT/MAX/MIN functions within python